Children's attributions of beliefs to humans and God: cross-cultural evidence

The capacity to attribute beliefs to others in order to understand action is one of the mainstays of human cognition. Yet it is debatable whether children attribute beliefs in the same way to all agents. In this paper, we present the results of a false-belief task concerning humans and God run with a sample of Maya children aged 4 to 7, and place them in the context of several psychological theories of cognitive development. Children were found to attribute beliefs in different ways to humans and God. The evidence also speaks to the debate concerning the universality and uniformity of the development of folk-psychological reasoning. Children’s Attributions 3 Children’s Attributions of Beliefs to Humans and God: Cross-Cultural Evidence Humans routinely attribute intentions, beliefs, and desires in order to interpret the behavior of others. Other humans are seen as agents, that is, as entities that pursue goals in accordance with their beliefs and desires. Attributions of agency are so ubiquitous that they are typically taken for granted in everyday life. These attributions are not always correct in identifying the beliefs and desires that underlie a specific action of an agent; yet, if people did not see others as agents, the capacity to understand their behavior would be severely impaired. For example, people would be surprised when others got up and moved. Abundant research documents children’s acquisition of human agent concepts over the first several years of life (Astington et al., 1988; Perner, 1993; Wellman, 1990), but there is little work available on the development of non-human agent concepts. Yet, people often attribute intentions, beliefs and desires to animals as well as to ghosts, gods, demons, and monsters. Scholars have long assumed that children first acquire concepts of human agency and then use them as templates to understand all nonhuman agents. One exception in this regard is found in the work of Barrett and collaborators (Barrett et al., 2001; Richert & Barrett, 2002; see also Atran, 2002, for an evolutionary account of why children cognize nonhuman agency). In this article, we offer further support for Barrett’s position, showing that Yukatek children do not reason in the same way about the agency of humans and God. In the first part, we discuss the development of human agent concepts, specifically with regards to the false-belief task. Then, we outline the predictions implied by several theoretical positions concerning the development of children’s understanding of humans’ and God’s beliefs. After that, we present experimental evidence from a Yukatek Maya sample to support the hypothesis that children do not reason about God strictly in human terms. Finally, we discuss the theoretical positions and Children’s Attributions 4 their predictions in light of the results and place the evidence in the larger context of theory of mind research. The Development of Human Agent Concepts The scientific literature on child development usually distinguishes three phases in the development of understanding of agency (see for example Csibra et al., 1999; Gergely et al., 1995; Gopnik & Meltzoff, 1997; Wellman, 1990). During the first year or so, children are believed to apply a principle of rational action – that is, they begin to appreciate that humans do not merely propel themselves, but do so in purposeful and rational ways. By the second or third year, children incorporate simple mentalistic attributions into this rational principle: the purposeful and rational action is understood to be driven by desires. Finally, during the fourth or fifth year, the principle of rational action is coupled with representational attributions: agents are seen as pursuing goals in accordance to their beliefs. The emergence of this representational stage, which is a necessary condition for attributing a full-fledged theory of mind to an entity, is the most relevant to the arguments and experimental results presented in this article. Exactly when this transition takes place has been a matter of considerable debate generating an abundance of research (e.g., Astington et al., 1988; Carruthers & Smith, 1996; Whiten, 1991). Although some evidence has emerged for the presence of representational reasoning in 3-year-olds (Chandler et al., 1989; Hala et al., 1991; Lewis & Osbourne, 1990; Siegal & Beattie, 1991), the bulk of the data available suggests that this ability is neither stable nor robust until children are five or older (Flavell et al., 1990; Perner et al., 1987; Wellman & Bartsch 1988; Wellman & Wooley 1990; Wimmer & Perner, 1983). Children’s Attributions 5 Since Premack and Woodruff (1978) started experimenting with non-human primates in order to establish the possibility that they had a ‘theory of mind,’ different ways of testing for this have been designed and tried out. As Dennett (1978) pointed out, you can credit an entity with a conception of belief only if there is evidence that it is able to understand that others may entertain false beliefs. Therefore, to probe children’s representational understanding of agency – whether they have a conception of belief – it is necessary to ascertain that they figure out that people can have false beliefs and that these beliefs can motivate behavior. A variety of false-belief tasks have been developed in the last twenty years to test children’s understanding of beliefs. One such experiment is known as the ‘Sally-Ann’ test (Wimmer & Perner, 1983). In this test, the child is made to look at a scene in which two dolls are animated by experimenters. The two dolls enter the stage; one of them (Sally) places an object in one of two containers and leaves the room. While Sally is out, the second doll (Ann) moves the object into the second container. Sally re-enters the stage; at that point, children are asked where Sally, who is unaware that the switch took place, will look for the object. Children, therefore, are asked to infer whether Sally will act according to her false belief (that the object is still in the original container) or not. Another false belief task, the one used in the experiments that will be presented later, is known as the ‘surprising contents’ task. In it, children are shown a closed container (usually a cracker box with a conspicuous picture of its contents on it) and asked what they believe is inside. The experimenter then opens the box to reveal that the crackers have been removed, and that small rocks (or a similarly unexpected item) have been put in their place. After reclosing the box, the experimenter checks that the children are still clear on what the box contains. The 1 The dolls are used to represent human beings. Wellman et al. (2001) have shown in a meta-analysis of false-belief studies that using a doll as a proxy for an actual human being does not affect the results of the task. Children’s Attributions 6 experimenter then introduces a doll who has not seen the inside of the box, and asks what the doll would think is in the container. Again, the point of the experiment is to establish whether children are capable of figuring out that other agents may have false beliefs and act accordingly. The Development of God Concepts By and large the research pertaining to children’s understanding of agent concepts deals exclusively with human agent concepts: how children’s concepts of human agency become increasingly specialized. In false-belief tasks, as well as in most other studies of children’s understanding of agency, experimenters have asked children to reason about human actions, beliefs, desires, and emotions. Very little available research addresses the generalizability of children’s understanding of agency to non-human agents in general, and to God in particular. However, by looking at the assumptions of several theoretical positions, we can envisage their predictions concerning the understanding of God in comparison to humans in a false-belief task. In Figure 1 we offer a tree diagram to show how these positions are related to each other. The graphs outline predictions of false-belief task performance in relation to humans and God. To illustrate the graphs, we will use a surprising contents task: suppose children are presented with a closed cracker box, shown that the box contains small rocks, and then asked what a human and God, who did not have a chance to look inside the box, would think is inside. Children’s Attributions 7 Figure 1. Predictions of False-belief Task Performance. 3/4 7 adult 3/4 7 adult